Error and occurrence analysis of Stanfins redesign at Computer Sciences Corporation

At Ball State University Dr. Wayne Zage and Professor Dolores Zage are working on a Design metrics project to develop a metrics approach for analyzing software design.The purpose of this thesis is to test the hypotheses of this metric by calculating the De external design component, and to show the correlation of errors and stress points in the design phase for a large Ada Software, professionally developed at Computer Sciences Corporation.From these studies we can relatively conclude that De does indicate the error-prone module. Since the D(G) is comprised of an internal and external component it is necessary to evaluate Di to support this hypothesis on a large project. Just by viewing the external complexity, the metric does a relatively good job of pointing out high error modules, with only viewing 10% of the modules we found 33% of the errors.Comparing the results of STANFINS-R and the results of the BSU projects, the BSU projects did better in finding the errors 33% verus 53%. However in the STANFINS project, we had a better success rate of finding the error modules. Of the modules highlighted 72% did contain errors. Thus if we loosened the criteria for selection of error prone modules we might have had a large percentage of the errors captured.Department of Computer ScienceThesis (M.S.